University of Central Florida University of Central Florida STARS STARS Electronic Theses and Dissertations, 2004-2019 2013 Governmental Responses To Terrorism: Creating Costs And Governmental Responses To Terrorism: Creating Costs And Benefits Benefits Kenneth Klose University of Central Florida Part of the International Relations Commons Find similar works at: https://stars.library.ucf.edu/etd University of Central Florida Libraries http://library.ucf.edu This Masters Thesis (Open Access) is brought to you for free and open access by STARS. It has been accepted for inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more information, please contact [email protected]. STARS Citation STARS Citation Klose, Kenneth, "Governmental Responses To Terrorism: Creating Costs And Benefits" (2013). Electronic Theses and Dissertations, 2004-2019. 2647. https://stars.library.ucf.edu/etd/2647 brought to you by CORE View metadata, citation and similar papers at core.ac.uk provided by University of Central Florida (UCF): STARS (Showcase of Text, Archives, Research &...
160
Embed
Governmental Responses To Terrorism: Creating Costs And ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
University of Central Florida University of Central Florida
STARS STARS
Electronic Theses and Dissertations, 2004-2019
2013
Governmental Responses To Terrorism: Creating Costs And Governmental Responses To Terrorism: Creating Costs And
Benefits Benefits
Kenneth Klose University of Central Florida
Part of the International Relations Commons
Find similar works at: https://stars.library.ucf.edu/etd
University of Central Florida Libraries http://library.ucf.edu
This Masters Thesis (Open Access) is brought to you for free and open access by STARS. It has been accepted for
inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more
successful when an adversary fears the imposed costs of retaliation. These costs can be
casualties, loss of equipment, and the opportunity costs associated with maintaining a military
1 Sageman (2008) and Gordon (2007) relate counterterrorism to containment in the sense that overt long-term conflict should be avoided while still opposing violent fundamentalism. Both share the belief that religious terror networks rely upon a moral indignation dependent upon a particular worldview as well as personal relationships. Terrorist attacks serve as public catalyst for inspiration and recruitment. If properly contained through denial and restriction, without governments resorting to provocative actions that could potentially legitimize terrorist causes, these movements may succumb to infighting (Gordan, 2007; Sageman, 2008).
15
force (Mearsheimer, 1995). There are numerous requirements for successful deterrence, such as
rationality and successfully communicating and understanding potential threats (Trager &
Zagorcheva, 2005/2006).2 These threats, whether implicit or explicit, must be understood and
believed according to estimated capabilities and political will (Trager & Zagorcheva,
2005/2006). The precision necessary to send and receive intended messages makes deterrence
difficult and potentially dangerous. If actors are incapable of formulating meaningful signals, and
recipents are unable to understand them with the intended meanings, then deterence may escalate
to unnecessary conflict (Jervis et al., 1985).
Many researchers explore whether deterrence is feasible considering terrorists are often
seen as irrational, willing to die, or difficult to find (Caplan, 2006; Kroenig and Pavel, 2000;
Sprinzak, 2000; Trager & Zagorcheva, 2005/2006). However, terrorist organizations, like all
organizations, assign responsibilities to individuals according to their roles: leaders plan, soldiers
execute attacks, recruiters replenish losses and garner support from surrounding populations,
external support from diasporas or even state sponsorship. Each of these components has
different characteristics and preferences that governments can exploit to achieve deterrence
Despite their reputations, high-level terrorist leaders rarely expose themselves to direct
action; targeting these leaders may prove effective to deter organizations (Bar, 2008; Fisher,
2007). Even individuals of supposedly unshakeable religious conviction can be deterred. Despite
2 Rational actors are assumed to gather information pertinent to the situation, assess possible actions according to the costs and benefits associated with implementation and finally select an action that has the largest benefit for the smallest effort (Morgan, 2003; Freedman, 2004; Kroenig & Pavel, 2012).
16
the public proclamation to fight until death, hundreds of Palestinian Islamic Jihad fighters
surrendered to Israeli forces in 2002 (Fisher, 2007). Even if leaders were too ideologically driven
to negotiate, the vast majority of individuals within their movements should still respond to
normal incentives (Caplan, 2006). While violence may have its purpose under certain situations,
overreliance upon violence can be dangerous possibly creating power vacuums, martyrdom, or
perceptions of indiscrimination due to inaccurate information regarding terrorist identities
(Heymann, 2002; Kalyvas, 2004).
The above literature review helped formulate the parameters of this investigation in
numerous ways. The long term causes of terrorism helped to identify control variables such as
urbanization, economic and political freedom, and economic development (Crenshaw, 1981;
Collier and Hoelffer, 2004; Bueno de Mesquita, 2005; Kavanangh, 2011).3 While the literature
on the short causes of terrorism helped formulate the hypotheses. For instance, conciliation may
actually increase terrorism in the short term due to spoiling attacks with potential long term
reductions due to improved intelligence from defectors (Bueno de Mesquita, 2005; Crenshaw,
1981; Gurr, 1998; Kydd & Walter, 2006). The responses proposed by Miller (2007) formed the
basis of the responses included in this investigation. Dugan & Chenoweth (2012) were
instrumental in shaping the method of investigation. Finally, a review of the literature shows
there are numerous datasets covering actual attacks but few resources devoted to governmental
responses. In order to address this absence, this project developed two entirely new datasets to
assess Algerian and Philippine responses to terrorism.
3 The investigation later had to discard political development due to a lack in variation in the selected time period, 2000-‐2010; however, the manner in which economic development was assessed included many important political elements such as corruption and government spending.
17
Description of Thesis Chapters
The next chapter discusses the theory, limitations and assumptions, as well as alterative
explanations. Since this investigation relies upon two newly created datasets, the entire third
chapter is devoted to the methodology for data collection and analysis. Chapters four and five are
quantitative assessments of governmental responses and their ability to reduce the overall
frequency and severity of terrorism in Algeria and the Philippines. They will also assess whether
these responses affected various terrorist groups differently. Both chapters will begin with
backgrounds to the respective conflicts within the countries. Chapter five will have a discussion
section that compares the results and discusses any resulting policy implications. Any identified
weaknesses of either the theory or the testing method will also be addressed. The chapter and the
thesis will conclude with propositions for future research.
18
CHAPTER TWO: THEORY
Introduction
This theory centers upon the interaction between three actors: terrorist organizations, the
government, and a potential support population for terrorism. Terrorists wish to change
governmental policy through violence applied to civilian populations. Governments must choose
a response and decide whether to apply it to the group or its leadership. The support population
must decide whether to support the terrorist or government.4
Definitions
Definitions for terrorism are almost as numerous as the recommendations to combat it.
Drawing upon a variety of sources, terrorism is premeditated politically motivated violence
against non-combatants in order to change governmental policy (Amos & Stolfi, 1982; LaFree &
Dugan, 2009; Sandler & Siqueria, 2006). Large political movements can produce these groups
when alternative peaceful means to achieve their political goals are unavailable (Crenshaw,
1981; Guevara, 1961; Gurr, 1998; Jones & Libicki, 2008). Terrorism is closely related to
insurgency, as both are methods of rebellion. Insurgents can use terrorism to create anarchy
within the state, weakening the incumbent government, making revolution more attainable
(Galula, 1964).
4 Likewise, Berman et al., (2012) have a model of insurgency with three actors: the rebel, the government, and civilians. Siqueira & Sandler's model (2006) of terrorism has three actors as well: a terrorist group, a government, and a terrorist support base. While the support population is an important component in the theory, unfortunately, the coding process did not reveal attempts by the the Algerian and Philippine governments to address the support populations. This issue will be discussed in the concluding chapter.
19
All three actors are assumed to be rational actors in the sense that they can prioritize
outcomes as well as select actions necessary to achieve goals (Caplan, 2006; Huth & Russet,
1984). Terrorism, particularly suicide terrorism, is seen as rational because it reflects a decision
that places more value on the fear created by an attack compared to the value an individual
brings to an organization (Pape, 2003). The very decision to engage in terrorism in the first place
represents a terrorist organization’s desire to achieve political change with the smallest amount
of resources through the maximization of limited assets (Betts, 2002). In addition, terrorists must
assume governments have the capacity to be rational; the government must be capable of
weighing continued violence against the costs of conciliation. The government’s ability to retain
power is usually contingent upon the support of its population (Buenos de Mesquita et al.,
2003).5 It is reasonable to assume that states prefer less frequent and less severe terrorist attacks,
as these attacks are direct challenges to the state’s ability to protect its citizens and faith in the
government. The support population is assumed to be rational as it must select not only who to
support but also the amount of its commitment, which is generally based upon a comparison
between the status quo and a potential future with more or less terrorism.
Actors
The Terrorist
The first actor, the terrorist organization, uses violence in pursuit of political goals
These attacks are designed to create fear not among the immediate victims but within a much
broader audience, those within the state that have enough influence with the government to
change policies (Miller, 2007).6
Terrorists attack to create fear yet their victims are often members of the very same
community it draws support. Because of this, terrorist must be mindful not to alienate the support
population. If the violence threatens the welfare of the support population, former allies may
become informants for the government (Galula, 1964; Gurr, 1998). To prevent defection,
terrorists threaten retaliation and/or provide competing services to those offered by the
government (Berman et al., 2012). Terrorist activity can be reduced by either a loss of capability
(material support, manpower, etc.) or the fear of losing necessary popular support, which
threatens their capabilities. Since terrorism requires a relatively small number of active
6 Buenos de Mesquita et al. (2003) refer to this segment of a state as the “winning coalition.” Who is a member of this group depends upon the internal politics of the state and often related to its regime type.
21
participants, at least relative to larger forms of rebellion such as insurgency or conventional
warfare, determining what is the necessary amount of popular support is problematic (Lawrence
2008; Sageman, 2008; Weinberg, 1991).
The Government
Governments seek to reduce levels of future violence and avoid concessions (Bueno de
Mesquita, 2005). It can be represented by an individual or agency involved in official activity
conducted to reduce terrorism ranging from law enforcement, political bodies, or military
personnel (Kennedy et al., 2012). The government seeks to reduce the frequency and severity of
attacks by convincing the terrorist organization that continued violence is not only harmful to
their physical well-being but also counterproductive to their goals. The government also seeks to
convince the support population (a community that the terrorist organization identifies with and
potentially draws support) that violence is not the optimal means by which to improve the status
quo and, therefore, to no longer support the terrorist organization through active or passive
means (Huth & Russett, 1984; Kennedy et al., 2012). Terrorists believe that with enough
violence, the government will change its policies. While the government desires less violence,
the amount it is willing to concede is dependent upon the severity and frequency of attacks as
well as the cost of complying with the terrorist demand (Dugan & Chenoweth, 2012; LaFree &
Dugan, 2009; Miller, 2007).
Governmental responses are often restricted. While enjoying an overwhelming advantage
in conventional power, its ability to use these advantages is curtailed by internal and
international norms regarding civil liberties and civilian casualties (Bar, 2008). Governments
22
choose responses available to them to increase the costs of terrorism. These increased costs deter
current terrorists while deterring members of the support population from joining or giving
passive support such as retaining information from the government (Fearon & Laitin 2003;
Kalyvas, 2009; Trager & Zagorcheva, 2006).
The government can choose from a variety of responses. This research design modifies
the list proposed by Miller (2007). It is believed using these responses will allow greater
variation to judge governmental responses more effectively. The following are the proposed
responses: conciliation, denial, legal reform restriction, and violence.
Conciliatory actions raise the cost of continued terrorism (Amos and Stolfi, 1982; Dugan
actions when counterterrorism campaigns will cost more than concessions. The government’s
core supporters must be willing to accept any proposed concessions. The amount the government
is willing to negotiate is dependent upon the severity and frequency of attacks as well as the cost
of complying with the terrorist demand (Dugan & Chenoweth, 2012; LaFree & Dugan, 2009;
Miller, 2007).
If the government chooses to make concessions through conciliatory actions, the terrorist
must decide whether to accept the offer or continue terrorism. Concessions can draw popular
support from terrorist groups if the offered concessions are viewed as sufficient. For instance,
Canadian concessions regarding Quebec autonomy successfully reduced terrorism levels (Lafee
and Dugan, 2009). If the terrorist chooses to accept the offer, the government may demand
cooperation from former terrorists. Those that continue to fight are most likely the extreme
elements of their movement. Ironically, although the government has created a schism in the
23
terrorist movement, the frequency and severity of attacks may actually increase in the short term
due to a more radical leadership and a desire to spoil negotiations. However, if the government
successfully obtains information from defectors, future governmental action will be more
effective causing long-term reductions in terrorism.7 Conciliation with leaders has the same
potential for spoiling attacks without the benefits of offering benefits to the majority of members.
The increase in frequency and severity should be even more pronounced.
Both sides fear the other’s commitment to any settlement. The government fears
continued violence and former terrorists withholding information. Former terrorist fear the
government abandoning its pledges following disarmament (Bueno de Mesquita, 2005). While
commitment issues are important they are simply beyond the scope of this research project.
H1A: Conciliation with groups will increase the frequency of terrorist attacks during a four-month period.
H1B: Conciliation with groups will increase the severity of terrorist attacks during a four-month period.
H1C: Conciliation with leaders will increase the frequency of terrorist attacks during a four-month period to a greater extent than conciliation with groups.
H1D: Conciliation with leaders will increase the severity of terrorist attacks during a four-month period to a greater extent than conciliation with groups
Acts of denial strengthen defenses such as installing metal detectors at airports or
Zagorcheva, 2005/2006). Even if improved defenses and safety measures decrease the frequency
7 It was impossible to statically assess the long term effects of governmental responses due to the scope of the investigation, 2000-‐2010. An annual unit of analysis would only allow 10 observations; therefore, a monthly unit of analysis was selected. Long term effects are however discussed in the final chapter.
24
of attacks, terrorists may simply shift their attacks to more vulnerable targets (Sprinzak 2000).
Rather than reducing terrorism, Enders & Sandler show that terrorists often shift to different, less
defended targets (1993). Following improved airport security after a series of airline hijackings
of the 1970s, terrorists shifted to less costly attacks such as assassinations (Sprinzak 2000). With
increased reconnaissance requirements arising from the need to find vulnerable targets, the
frequency of attacks may decrease. However, due to more lengthy planning periods and the
concentration of resources, attacks may become more effective, resulting in an increase in
severity.
H2A: Denial will reduce the frequency of terrorist attacks during a four-month period.
H2B: Denial will increase the severity of terrorist attacks during a four-month period.
Legal restrictions are enacted to increase the government’s abilities to counter terrorism.
If the government has strong support among the general population, it may consider restriction
through legal reform. If, popular support is lacking or if the government is an authoritarian
regime, then it may bypass legislative reforms and simply declare emergency powers.
Regardless, the result is enhanced governmental authority and power. Legal restriction is similar
to denial, while the latter focuses on the physical barriers to mobility, legal restriction focuses on
authority. Both intend to reduce the ease of launching attacks; however, they do little to reduce
the capacity to launch attacks in the short term. The reduced frequency of attacks may have the
unintended consequence of forcing terrorists to pool resources resulting in an increase in the
severity of attacks.
H3A: Legal restriction will decrease the frequency of terrorist attacks during a four-month period.
25
H3B: Legal restriction will increase the severity of terrorist attacks during a four-month period.
Violence is when the government attempts to kill or capture terrorists through a wide
range of activities such as targeted assassinations or raids (Davis & Jenkins, 2002; Kroenig &
isolation, violence may alienate members of the support population. Terrorist groups will seek to
demonstrate their resolve to both the government and the support population following even
successful violent operations. However, due to violent governmental action, their capabilities
may be reduced (Posen 2001/2002). Follow on attacks may be conducted in a hasty fashion,
lacking proper assets and coordination (Byman, 2006). Violence against leaders will reduce both
the frequency and severity of attacks due to the organizations reduced planning and coordination
capabilities.
H4A: Violence towards groups will increase the frequency of attacks during a four-month period.
H4B: Violence towards groups will decrease the severity of terrorist attacks during a four-month period.
H4C: Violence targeting leaders will decrease the frequency of terrorist attacks during a four-month period.
H4D: Violence targeting leaders will decrease the severity of terrorist attacks during a four-month period.
Alternative Explanations
This research makes no distinction between domestic and transnational terrorism or the
pragmatism of particular terrorist goals (temporal/transformational goals). While perhaps
important distinctions, they are simply beyond the scope of this research project. The difference
26
between temporal and transformational goals may be more important than the specific type of
ideology. For instance, there may be important differences between a religious group that wants
to install Sharia law in its own country and another religious group that wants to create an
Islamic caliphate across the Middle East (Davis & Jenkins, 2002; Richardson, 2007).
Considerable effort was made to differentiate governmental responses as opposed to
solely using broader categories such conciliation and repression (Dugan & Chenoweth 2012;
Lyall 2009). However, using these specific responses may be no more insightful considering a
response’s scale is still not taken into account. For instance, a government could conduct military
operations with 100 or 1000 personnel. Resource allocation could send powerful commitment
signals to both terrorists and support populations.8
Even if the government is effective in creating costs and benefits, various types of
terrorist organizations may react differently based upon their ideology or degree of support from
diasporas and/or state sponsors. External support may reduce the government’s ability to impose
costs upon terrorist organizations or support populations. With considerable outside support,
terrorist organizations may be less responsive to the concerns of the support population, reducing
the effects of conciliatory actions regardless of how or to whom they are applied (Siqueira &
Sandler, 2006).9
8 An attempt was made to assess the scale of large troop movements using the log of deployments of troops beyond company level. They were omitted due to higher collinearity with other manifestations of violence such as arrests and the killing of terrorist suspects. 9 Collier & Hoeffler (2004) use the number of foreign immigrants in the United States as a proxy for diasporas support; however, this was deemed unsatisfactory since there is no way to determine if the minority group actually supported terrorism. An attempt was also made to control for foreign government support using the BAAD-1 database and Terrorist
27
Key Assumptions and Limitations
It is assumed all three actors are rational. This is essential. The terrorist must be able to
measure costs imposed by governmental action such as violence against its leadership.
Governments must be able to weigh the costs and benefits of conciliatory actions particularly
when domestic populations have the means to punish the government for unpopular actions as in
elections. Governments must also consider how other terrorist groups will react to government
concessions to a particular movement or terrorist organization. Support populations must be able
to weigh the costs of defecting while considering whether the state is acting discriminately or
indiscriminately.
In addition, since the theory is dependent upon cost imposition, it may not be appropriate
for lone wolf terrorists. This may become increasing problematic as instances of lone wolf
terrorism increase. Examples include the Oklahoma City bombing in 1995 and Theodore
Kaczynski, but also increasingly with self-radicalized Islamic terrorists, such as when
Abdulhakim Mujahid Muhammad attacked an army recruiting station in Little Rock, Arkansas
(Carter & Carter, 2012).
The case selection may create data availability issues. The wire services may not
adequately capture governmental actions in remote areas. It may also arise from language
barriers. This possibility is reduced for the Philippines as English is one of its official languages,
while Arabic is Algeria’s official language.10
Organization Profile database; however, due to conflicting reports within these datasets and other resources, the degree of foreign support was omitted. 10 While not an official language, French is widely used by government officials.
28
Contributions to Literature
Rather than simply deciding between broad responses, this research design investigates a
series of potential responses. In addition, few studies that address specific responses adequately
control for economic and political factors not to mention terrorist ideologies. While not specified
as control variables, terrorist ideology was accounted for by investigating not just the frequency
or severity of terrorism at the national level but also by individual terrorist groups reflecting
religious as well as revolutionary groups. Additionally by investigating the effects of
governmental responses on both the frequency and severity of terrorism, we will hopefully gain a
more nuanced understanding of the unintended consequences of counterterrorism policy. Lastly,
by differentiating between actions that target groups or leaders, we can assess various means to
impose the greatest benefits to those that forgo terrorism while imposing the greatest costs to
irreconcilable terrorists.
29
CHAPTER THREE: METHDOLOGY
Introduction
Using time series regression, governmental responses to terrorist attacks between 2000-
2010 in Algeria and the Philippines will be evaluated according to their ability to reduce both the
frequency and severity of attacks while controlling for economic and political factors. The
period, 2000-2010, was selected in order to have lagged variables with the available economic
and political datasets.11 Data was collected from September 1999 to December 2010 to allow for
lagged variables.
Case Selection
Algeria and the Philippines were selected to allow for the greatest amount of
generalizability despite having two cases. The Philippines is a democracy while Algeria is an
anocracy according to Polity IV.12 This difference may show certain regimes are more prone to
different responses. It may also show that there are variations in how regimes react to different
types of terrorist groups. Both allow for religious variation. The majority of Algerians are
Muslim while in the Philippines, the majority are Roman Catholics. While not a focus of this
project, it will be interesting to see if religious attributes have an impact upon governmental
responses. Will Algeria, a majority Islamic country, react to religious groups the same way the
11 The Heritage Index of Economic Freedom has data available from 1995-‐2013, while the World Bank and Polity IV datasets have extensive amounts of data for decades. The POLITY IV dataset as a control variable was omitted due to a lack in data variation. 12 Polity IV has considered the Philippines a democracy following the popular overthrow of President Marcos in 1987 (Polity IV Country Report 2010: Philippines).
30
Philippines reacts considering their majority Roman Catholic populations? Ethnically, Algeria is
nearly homogenous while the Philippines is highly heterogeneous with no ethnic group
comprising more than 30 percent of either population (CIA Factbook).13
Between 2000 and 2010, both Algeria and the Philippines suffered over 500 terrorist
attacks (Global Terrorism Database).14 According to the Global Terrorism Index, Algeria and the
Philippines have been consistently in the top 10 states affected by terrorism. In addition, the
cases were selected in order to allow variation in the types of terrorist organizations; each have a
range of terrorist organizations including revolutionary and religious groups.15
Variables
Dependent Variables
The dependent variables, the frequency (total number of incidents in a month) and
severity (total casualties including wounded and killed in a month) of attacks were obtained from
the Global Terrorist Database (GTD) produced by the Study of Terrorism and Responses to
Terrorism (START) at the University of Maryland. In order to restrict the data to terrorism rather
than criminal activity or insurgency several criteria were used. The attacks must have been in
support of a political goal, there must have been an attempt to coerce a larger audience, and they
must be outside the bounds of legitimate warfare, such as the intentional targeting civilians or
13 N i n e t y n i n e percent of Algerians are Arab-Berbers; however, only fifteen percent self-dentity as Berbers (CIA Factbook). 14 These figures used the most restrictive criteria to ensure the data truly reflected terrorist activity as opposed to insurgency. Without these restrictions, Algeria and the Philippines have over 1200 attacks according to the GTD. 15 Unfortunately, the Moro National Liberation Front, a national separatist group, had to be omitted due to a lack of activity between 2000-- 2010.
31
non-combatants. All ambiguous cases were excluded. All unsuccessful attacks were included.
The aforementioned criteria are options provided by the GTD.
The frequency and severity of attacks in Algeria and the Philippines were aggregated on a
monthly basis by country as well as by individual terrorist groups from 2000 to 2010. These
distinctions were important to see if governmental responses had similar effects across a variety
of terrorist organizations. In Algeria, the investigation included Al-Qaeda in the Islamic Maghreb
(AQIM) and the Armed Islamic Groups (GIA). Since AQIM and the Salafist Group for Call and
Combat (GSPC) are essentially the same organiation, the frequency and severity of both
organizations were combined under AQIM. The background section will discuss this relationship
in detail. In the Philippines, the investigation included Abu Sayyaf (ASG), Jemaah Islamiyah
(JI), the Moro Islamic Liberation Front (MILF), and the New People's Army (NPA). Attacks by
unknown perpetrators were also individually assessed for both countries. Terrorist groups with
three or less attacks between 2000 and 2010, according to the GTD, were included in the overall
assessments however excluded from individual investigation. For a summary of the frequency
and severity, see Table 1: Frequency and Severity of Terrorism by Country and Perpetrator
32
Table 1: Frequency and Severity of Terrorism by Country and Perpetrator
Note: The unit of analysis was on a monthly basis however for the purposes to presentation the data was aggregated to the annual level.
Independent Variable
Governmental responses were categorized as one of the following responses:
conciliation, denial, legal restriction, or violence. Unless otherwise specified, these variables
were coded as a numeric value for the number of such actions performed in any given month.
Each governmental response was assessed with 0-3 month lags to see if effects fluctuated
33
over time.
Conciliation is when the government attempts to reach a compromise or political
settlement such as offering amnesty programs, signing ceasefire agreements, or the release of
prisoners (Amos and Stolfi 1982; Dugan and Chenoweth 2012; LaFree & Dugan, 2009; Miller,
2007). Acts of conciliation were separated between those that targeted the group at large and
those targeting leaders.16 For instance, the Philippines offered amnesty to members of the New
People’s Army (NPA) in 2007. Conversely it offered protections from prosecutions for NPA
leaders that participated in negotiations with the government. An act of conciliation was coded
as 1 if present, and 2 if it was accompanied by a significant prisoner release.
Denial was coded as an attempt to restrict physical movement either of supplies or
manpower to conduct terrorist attacks. Denial consists of the use of roadblocks, curfews, or the
installation of cameras in key locations (Adams 2003, Freedman, 2004; Heymann 2001/2002;
Knopf, 2010, Miller, 2007, Trager & Zagorcheva, 2005/2006). While both Algeria and the
Philippines are highly militarized countries where military or police checkpoints are a part of
daily life, only the incidents of roadblocks as a broad regional or national policy were positively
coded. Denial was coded as a 1 if any of the above actions took place in that given month.
Legal restriction consists of instances when the government enacted laws or changed
national policy to increase the state’s law enforcement abilities or restrict the media’s coverage
of terrorism. For instance, in 2007 the Philippines enacted the Human Security Act, which allows
the detention of suspects for three days without a warrant (BBC, 2007). Legal restriction w a s
16 To be considered a leader, the individual had to mentioned by name, be within the top three of the organization’s hierarchy, or listed as a key leader in its Terrorist Organization Profile (Memorial Institute for the Prevention of Terrorism, 2006).
34
coded as a 1 if a law or policy took effect that strengthened its legal powers or if it declared
emergency powers.
Violence are instances when the government killed or captured terrorists (Davis &
from Factiva after searching for “Israel” (2012).17 The resulting information was used to
compile the Government Actions in a Terror Environment – Israel database (GATE-ISRAEL).
To collect articles, searches were conducted for “Algeria” and “Philippines” using
Factiva. The searches were restricted to Reuters, Agence France Presse, and the Associated
Press. Rather than using a filter program as used by Dugan and Chenoweth (2012), the
researcher used built-in filter options in Factiva to capture only articles that pertained to
governmental responses to terrorism. These filters reduced the number of articles from 56,325
and 155,497 to 4,239 (Algeria) and 17, 495 (the Philippines). For a summary of the Factiva
filters, see Table 2: Selection Filters.
17 Factiva is an online service provided by Reuters and the Dow Jones that allows access to thousands of sources from dozens of countries in numerous languages.
35
Table 2: Selection Filters
Key Word Date Source Subject Region Language Algeria Sept 1999-Dec
2010 AP, AFP, Reuters
National security, risk news, excluding disasters
Algeria English
Philippines Sept 1999-Dec 2010
AP, AFP, Reuters
National security, risk news, excluding disasters
Philippines English
In order to code these articles, the process used NVivo, a text analysis program, rather
than TABARI. While TABARI is free and highly manipulatable, it also requires programming
experience. In contrast, NVivo is user friendly, and while it is a commercial product, many
universities make it available at their libraries and computer labs.18 Using NVivo, the selected
articles underwent a five stage filtering and coding process to screen for specific governmental
responses. During stages 1-3, a search was conducted for a series of key words. Unlike many
other programs, Nvivo allows the researcher to search for stemmed words, synonyms, or exact
phrases.19 During stage 4, each observation was individually evaluated to ensure correct coding.
Stage 5 corrected for duplicate observations and recorded the results in separate datasets for
Algeria and the Philippines. Each dataset consists of categories of data pertaining to
governmental action including offers of amnesty, arrest of terrorism suspects, the use of heavy
weapons (artillery, air strikes), signed ceasefire agreements, denial, legal restriction, loss of
terrorist leaders due to governmental action, referendums and/or elections, release of prisoners,
18 NVivo is available at the University of Central Florida at the Graduate Student Center, Coburn Hall, Suite 146. 19 The quality of results was not uniform. It required trial and error in order to select the proper level of discrimination (stemmed words, synonyms, or exact phrases)
36
and major troop movements. The collected data was in excess of the variables investigated in this
particular research project; however, it may prove invaluable for potential future research.20 For
a summary of the filtering process, see Table 3: Filtering Process to Obtain Governmental
Responses.
20 The codebook for Government Actions in a Terrorist Environment – Israel (GATE-Israel) and the Measures against Extremism and Terrorism (CoMET) Database provide guidelines for key terms for coding.
37
Table 3: Filtering Process to Obtain Governmental Responses
Steps Goal Method Key Words Dis. Notes
Step 1 Filter for gov. actors
“army OR authority OR congress OR country defense OR defence OR government OR intelligence OR law enforcement OR military OR nation OR national OR officials OR parliament OR police OR president OR prime minister OR security OR soldiers OR state.”
Stemmed
Step 2
Filter for gov. actions against terrorists
"militant OR militants OR extremist OR extremists OR fundamentalist OR fundamentalists OR separatist OR separatists OR reactionary OR reactionaries OR terrorist OR terrorists OR Marxists OR revolutionary OR revolutionaries OR communist OR communists OR Salafist OR Salafists OR Islamic OR Islamics OR Al-Qaeda OR rebel OR rebels"
Synonyms
Filter for Con.
"accord OR agreement OR amnesty OR autonomy OR cease-fire OR cease fire OR compensation OR conciliation OR concord OR exchange OR free OR freedom OR leader OR negotiate OR pardon OR peace OR reconcile OR reconciliation OR reform OR released OR referendum OR resolution OR transfer OR withdrawal OR surrender"
Stemmed
Filter for Denial
"curfew train OR railway OR airports OR barriers OR deny OR defense OR defence OR metal detect OR guards OR protect OR patrol OR reinforce OR roadblocks OR road blocks OR checkpoint OR roadblock OR tightened OR security measures"
Stemmed
Filter for Legal Restrict.
"ban OR congress OR convict OR law OR law enforcement OR legal OR police OR legislation OR parliament OR resolution OR restriction OR gendarme OR gendarmerie OR checkpoint OR roadblock"
Maleckova, 2003). In order to increase variation, the original data was converted from annual
data to the change of GDP growth from year to year.
In order to control for economic freedom, the study used The Heritage Index of
Economic Freedom, a composite score.21 Economic freedom at face value may not appear to
contribute to political violence. But considering how the self-immolation of Mohamed Bouazizi
following the confiscation of his goods sparked protest across the Middle East, in what became
known as the Arab Spring, it becomes apparent how economic freedom can potentially affect
levels of political violence (Al Jazeera, 2012).
To control for changes in demographics and population density, the population (in
100,000) and the percentage of urbanization were included. Growing populations can be more
difficult for governments to control as well as increase the recruitment pool (Fearon et al., 2007;
21 The overall score is derived from four main categories: “Rule of Law (property rights, freedom from corruption), Limited Government (fiscal freedom, government spending), Regulatory Efficiency (business freedom, labor freedom, monetary freedom, and Open Markets (trade freedom, investment freedom, financial freedom) (Heritage Index of Economic Freedom).”
39
Fearon & Laitin, 2001; Sambanis, 2007). Urbanization can potentially give terrorists increased
mobility, communication, and access to targets. While insurgencies may benefit from rural or
mountainous terrain, terrorism on the other hand may be better facilitated by urbanization
Maleckova, 2003). The negative effects of unemployment should also be more pronounced
among highly educated individuals as increased education raises expectations. The percentage of
unemployed males with higher education was considered but later omitted due to data
availability issues with Algeria.
While the effects of political freedom are often discussed in terrorism and insurgency
literature, both Algeria and the Philippines experienced little change in political development
between 2000-2010. While individual scores may have changed over the course of several years,
the overall assessments for both countries remained the same, Algeria remained an anocracy
while the Philippines remained democratic. Political freedom was therefore excluded.
Lastly, since both governments faced religiously inspired terrorist groups (AQIM, GIA,
ASG, JI, MILF), it was important to control for events of Islamic significance such as Ramadan.
22 Due to a gap in unemployment data from Algeria, unemployment rates were averaged using the preceding and following years. The average unemployment rate among males was calculated for the years 1998, 1999, 2001, and 2002. Records from the Algeria’s Office National des Statuesque’s were consulted however the gaps persisted.
40
This was not initially a consideration; however, during the coding process numerous articles
mentioned the government’s need to increase security because of the approaching holiday.
Methods
The assessment of governmental responses to terrorism was conducted using a Feasible
GLS, Prais-Winsten time series regression. All the variables were tested for stationarity using the
augmented Dickey Fuller test. In order to standardize the data, all variables were treated with
first differencing. Another round of augmented Dickey Fuller tests were conducted to ensure the
data was successfully rendered stationary. In instances when the results were inconclusive, a
modified Dickey Fuller (DFGLS) test was applied. After the data was sufficiently deemed
stationary, time series regressions were conducted. After the regressions, alternative Durbin and
Breush-Godfrey methods tested for serial correlation. The Prais-Winsten method was used to
correct for serial correlation. The Prais-Winsten method was selected as opposed to the
Cochrane-Orcutt regression due to how the latter drops the initial observation. The above
methods were applied to both Algeria and the Philippines, aggregated at the national level as
well as for individual terrorist organizations.
The theory is tested using three models. All models include governmental responses
(conciliation with groups, conciliation with leaders, denial, legal restriction, violence against
groups, and violence against leaders) and the controls (economic freedom, annual change of
GDP, population, Ramadan, and percent of urbanization). All models include governmental
responses between 2000-2010. All controls except for Ramadan were lagged by 12 months.
Models 1 and 2 are restricted to overall levels of frequency and severity at national levels. Model
1 lags governmental responses by 1 month. Model 2 lags governmental responses by 3 months.
41
This is done to isolate any temporal effects. Model 3 is far more comprehensive. In
addition to overall levels of frequency and severity, Model 3 also investigates whether
governmental responses have different consequences when applied to different groups.
In Algeria, the investigation includes Al-Qaeda in the Islamic Maghreb (AQIM), the
Armed Islamic Groups (GIA), and unknown attackers (UNK). In the Philippines, the
investigation includes Abu Sayyaf (ASG), Jemaah Islamiyah (JI), the Moro Islamic
Liberation Front (MILF), the New People's Army (NPA), and unknown attackers.
Model 3 also includes the initial month governmental responses took place as well as
three lags, 1 month, 2 month, and 3 month lags.
All responses taken by a government between 2000-2010, were used for overall
levels of frequency and severity. When investigating an individual group, only
responses that were directed towards that specific group were used, unless specifically
identified. For a summary of the models, see Table 4: Model Summary. All regressions
are presented in their entirety in the appendixes. Any deviations from the models are
acknowledged and used mainly as discussion points and not part of the formal
regressions.
Table 4: Model Summary
42
CHAPTER FOUR: ALGERIA
Introduction
Using data from 2000-2010, the Algerian government’s responses to terrorism were
assessed using time series regression. The first section of the chapter provides a brief background
to the Algerian conflict. The results of the analysis are then presented. Models 1 and 2 will be
discussed at the same time while focusing on areas of statistical significance. Since Model 3 is
far more comprehensive, it will be discussed in detail while addressing each hypothesis
individually.
All the models include the same dependent variables, frequency and severity as well as
the same governmental responses, and controls. Models 1 and 2 only investigate overall levels of
frequency and severity. Model 1 includes a 1 month lag. Model 2 includes a 3 month lag. In
contrast, Model 3 also investigates effects upon individual groups. Model 3 includes multiple
lags to include 1, 2, and 3 month lags as well as the initial month.
Given the ten-year scope of the investigation, there was not enough observations to assess
responses on an annual basis. However, potential long term effects, particularly those of
conciliation and violence, are discussed in the final chapter. Discussion points, final comments,
gaps in the theory, and potential for future research will be discussed in the final chapter.
43
Background
For decades, violence has marred Algerian history. Algeria was colonized by France in
1830 and remained a firm part of the French empire until the conclusion of the Second World
War and subsequent break up of overseas European empires. From 1954 – 1962, Algeria fought
a war of independence using both insurgency and terrorism, costing the lives of over a million
Algerians. (BBC, 2013; Shugart, 2006). Following a military coup in 1965, Algeria emerged
constitutionally committed to socialism with the National Liberation Front (FLN) as the sole
inheritor of power and authority (Home Office, 2013).
Over the next ten years, Algeria experienced relative stability, if at the expense of
political development. Unfortunately in 1989, an experiment in democracy ended in disaster.
After allowing increased political participation, dozens of parties contested the parliamentary
elections in June 1991. The newly formed Islamic Salvation Front (FIS) won 188 seats in the
first round and more success expected in the second round. Faced with potentially losing his
majority, President Chadli dissolved the National People’s Assembly (Oberschall 2004; Home
Office, 2013; International Crisis Group, 2001). Following his subsequent resignation, a five-
member council headed by Mohammed Boudiaf effectively ruled Algeria. The FIS was banned
and a state of emergency declared. Despite its intended length of one year, it persisted for over
19 years (Lowe, 2011). Later in 1996, all religious parties were banned (Home Office, 2013).
The assassination of President Boudiaf forecasted a decade of insurgency and over twenty years
of terrorism (International Crisis Group, 2001).
During the first years of the conflict, the Algerian government and the FIS attempted to
negotiate a compromise; however, the conflict quickly entered a downward spiral with increased
44
levels of indiscriminate violence perpetrated by both sides. Frustrated by a lack of progress,
many FIS supporters defected to the more radical splinter organization, the Armed Islamic Group
(GIA). From its inception, the GIA used indiscriminate violence against civilians, particularly
foreigners (Memorial Institute for the Prevention of Terrorism, 2008). Killing civilians at fake
checkpoints quickly became one of its favorite tactics (Reuters, 2005; AP 2009). Facing a drain
of support, the FIS created its own armed wing, the Islamic Salvation Army (AIS) with the
intention of creating an Islamic state under Sharia law (Memorial Institute for the Prevention of
Terrorism, 2008).
Since 1992, over 150,000 Algerians have been killed with both sides accusing the other
of brutal indiscriminate violence (BBC, 2013; Home Office, 2013; Reuters, 2005). The Algerian
government has accused the GIA of killing entire villages, while the GIA has made counter
accusations that the massacres were attempts to discredit their legitimate political grievances
(Cronin 2006; Renard, 2008). As the war escalated Algerian security forces were increasingly
accused of vigilante violence, sometimes even at random, due to increased desperation and
frustration (International Crisis Group, 2001).
Between 1995-1998, the government began major offensives nearly destroying the AIS
as an effective fighting force, which placed renewed pressure upon the FIS to negotiate. In 1997,
the FIS declared a unilateral ceasefire (Memorial Institute for the Prevention of Terrorism,
2008). In 1999, Abdelaziz Boutflika won his first Presidential elections (BBC, 2009). Following
a series of tactical victories, Boutflika proposed an amnesty for referendum approval. In
September 1999, a vast majority of Algerians approved the Civil Concord, an amnesty program
for those not guilty of crimes against civilians. To the dismay of victims and their families,
45
President Boutflika extended the amnesty by presidential decree to include accused terrorists, as
long as individuals turned themselves in before January 13, 2001 (BBC, 2013; International
Crisis Group, 2001).
After the government promised to remove restrictions on political participation and
release political prisoners, the AIS disbanded in January 2000. An estimated 80 percent of the
insurgents accepted amnesty (BBC, 2013; Home Office, 2013). Between September and
December 1999, thousands of political prisoners were released. Despite promises to extend
political rights to all Algerians, the FIS remained banned as did its successor, the Wafa party
(International Crisis Group, 2001).
While the amnesty neutralized thousands of insurgents and caused the dissolution of AIS,
it had the unintended consequence of consolidating violent opposition in increasingly radical
hands. The main beneficiaries of remaining fighters were the GIA and the Salafist Group for Call
and Combat (GSPC), which separated from the GIA in 1998. The remaining groups share the
same organizational and ideological linage reflecting a continued process of splintering with
increased radicalization at every separation (Memorial Institute for the Prevention of Terrorism,
2008).
Facing a smaller yet just as lethal terrorist threat, Algerian security forces reportedly used
questionable methods. Things deteriorated to the extent that even news wire services commented
on the infrequent arrests relative to the frequent deaths of terrorist suspects. According to a report
disclosed by a government appointed committee in March 2005, security forces were responsible
for the unlawful disappearances of over 6,000 individuals. Despite accusations of excessive
violence, Algerian security forces continued to make tactical gains, such as the arrest of GIA
46
leader Nourredine Boudiafi in January 2005. In times of increased threat levels, significant
resources were used to establish a ring of roadblocks around Algiers, which successfully
prevented a terrorist attack within the city for nearly two years although violence continued
elsewhere (AFP, 2010).
Experiencing similar tactical gains as in 1999, Boutflika once again offered an amnesty
for approval by referendum in September 2005. An estimated 97 percent of Algerians approved a
six-month amnesty that began in March 2006. Under the Charter for Peace and National
Reconciliation, both rebels and security forces were pardoned. Thousands of political prisoners
were released, including the founder of the GIA, Abdel-Haq el-Ayadia (Home Office, 2013).
Diminished by military operations and the multiple amnesties, GIA members reportedly
began defecting to its successor, the GSPC (Memorial Institute for the Prevention of Terrorism,
2006). It appears as though tactical and political success left a smaller yet equally lethal terrorist
threat. After becoming the eminent terrorist threat in Algeria, the GSPC began modifying its
targeting, tactics, and propaganda. The group increasingly targeted softer foreign commercial
targets using explosives rather than firearms reflecting a shift from insurgency to terrorism. In
2003, it announced its alliance with Al-Qaeda. Four years later, it officially changed its name to
the Al-Qaeda Organization in the Islamic Maghreb (Memorial Institute for the Prevention of
Terrorism, 2006; Loidolt, 2011; Render, 2008). To visualize how the frequency of GIA and
AQIM changed, see Figure 1: Comparing GIA and AQIM Frequencies of Terrorism.
47
Figure 1: Comparing GIA and AQIM Frequencies of Terrorism
As seen with the recent deaths of 37 hostages in 2013 at an oil facility, a joint
Algerian-European business venture, radical extremism remains a threat both to Algeria and
foreign business interests in the country (CNN, 2013). However, the frequency and severity of
attacks has been reduced significantly. The following section will use time series regression to
attempt to assess the government’s responses.
Descriptive Statistics
Table 4 provides descriptive statistics on a monthly basis from January 2000 to
December 2010. The dependent variables, the frequency and severity of terrorism are also
02
46
8N
umbe
r of A
ttack
s
Jan 2000 Jan 2005 Jan 2010mm_yr
Frequency - GIA Frequency - AQIM
48
included. Data from September 1999 to December 1999 was excluded since it was collected to
create lags.
Table 5: Descriptive Statistics for Governmental Responses, Algeria
vio_group_unk 10.197 21.593 0.000 149.000 Note: The above figures only include data from 2000-2010. Data was collected to allow for lags between September 1999 and December 1999. Additionally, denial and legal restriction are not broken down by group since Algeria applied these responses equally to all groups. (n = 132).
Overall, there were an average of 4 attacks a month, resulting in an average 29 casualties.
AQIM averaged of 1 attack and over 10 casualties a month.23 When GSPC changed its name to
AQIM, it affected both its frequency and severity of attacks, reflecting changes beyond rhetoric
but a significant increase in lethality (See Figure 2: Overall Frequency and Severity of Terrorism,
23 AQIM was treated as the same organization for averages and regressions. When specifically discussing averages prior to its name change in 2007, it will be referred to as GSPC.
49
Algeria). Prior to February 2007, the GSPC averaged less than 1 attack a month, doubling to
almost 2 attacks after becoming AQIM. The subsequent increase in severity was even more
dramatic, increasing from an average of 3 monthly casualties to over 22 (See Table 6: Effects of
GSPC transition to AQIM, Algeria). AQIM’s deadliest month also occurred during the same
period.
Figure 2: Overall Frequency and Severity of Terrorism, Algeria
50
Table 6: Effects of GSPC Transition to AQIM, Algeria
Organization Mean SD Min Max Obs.
GSPC - Frequency 0.8453 0.9740 0 4 85 AQIM - Frequency 1.9787 1.6217 0 7 47
GIA averages are underestimated as it was most active between 2000-2003, with only
four attacks between 2004-2010. At the height of its activity, the GIA averaged 1 attack and 11
casualties a month. Unknown perpetrators were the most active and lethal, averaging 2 attacks
and 14 casualties a month over the 132-month period. This was only surpassed by AQIM
following its international transition. Between 2000 and 2003, the unknown attacks were most
likely split evenly between the GIA and GSPC. This is based on the known attacks in that same
period. Most of the casualties, however, were most likely caused by the GIA as it was the most
lethal group during that time.
Table 7: Active GIA Period, January 2000 to December 2003, Algeria
Organization Mean SD Min Max Obs. GIA - Frequency 1.1875 1.4241 0 6 48 GIA – Severity 11.5417 17.6031 0 66 48
On average, Algerian security forces killed or arrested 18 terrorists a month during the
132-month period. Breaking down the use of violence against groups, security forces killed or
arrested 4 members of AQIM and 2 members of the GIA a month. Unfortunately, the articles
often contained too little information to assign a group’s membership to an average of 10 arrests
51
or killing of terrorists a month. To see if security forces targeted specific groups more frequently
at various periods, it is possible to analyze the averages for specific periods such the GIA most
active period as well as the GSPC’s transition to AQIM. It appears that Algerian security forces
targeted specific groups by their levels of lethality. During the height of the GIA’s activity,
between 2000 and 2003, Algerian security forces killed or arrested an average of 7 members,
dropping to less than 1 from 2004 to 2010. During the same period, security forces only killed or
arrested an average of 2 members of GSPC; however, the use of violence dramatically increased
to an average of 7 following the GSPC transition to AQIM. This may also reflect diminished
support from the population, particularly if AQIM was seen as not representing the best interest
of Algerians due to its increased internationalization.
Table 8: Governmental Use of Violence, Algeria
Organization Period Mean SD Min Max Obs. GIA - Active Jan 2000 - Dec 2003 7.2917 17.7716 0 110 48 GIA - Inactive Dec 2003 - Dec 2010 0.0595 0.4486 0 4 84 GSPC Jan 2000 – Dec 2003 2.3333 6.59249 0 33 48 GSPC Jan 2000 - Feb 2007 2.9419 6.5769 0 33 86 AQIM Feb 2007 - Dec 2010 7.6170 13.5752 0 48 47
Results
The results are mixed requiring an understanding of the particular circumstances in
Algeria during the period of investigation. The results for the overall levels of frequency and
severity at the national level as well as for specific terrorist groups were achieved using separate
regressions. Models 1 and 2 will be discussed together while focusing on areas of statistical
significance. Frequency will be discussed first followed by severity. The regressions can also be
found in Appendix A (Model 1 and 2). Results for Model 3 tended to have greater statistical
significance; as a result, Model 3 will be discussed by itself while addressing individual
hypotheses.. Model 3 will be discussed by itself while addressing individual hypotheses. In order
to facilitate the review of multiple lags, specific results were taken from separate regressions to
form the tables. The original regressions for Algeria can be found in Appendix C. All regressions
used robust standard errors.
Models 1 and 2Frequency
The results do not show statistical support for H1A, H2A, H3A, H4A (See Table 9:
Models 1 and 2: Effect of Gov. Responses on the Freq of Terrorism, Overall, Algeria). However,
there was support for H1C and H4C with Model 1 and its 1 month lag. In support of H1C,
conciliation with leaders was significant at the 1% level with a coefficient of 6.383. Algeria
averaged 4.197 attacks a month. As a result, conciliation with leaders would increase the
frequency of attacks by nearly 2 attacks a month. In support of H4C, violence against leaders
had a coefficient of -1.907 with a 10% significance level. Once again, considering that Algeria
averaged 4.197 attacks a month, this would represent nearly a 50% reduction.
Model 2 with a 3 month lag had results in opposition to H4A; however, the real life
implications were minor. Violence against groups led to a decrease of -.026 attacks with a
significance level of 5%. While in opposition to the theory, the real life effects would be marginal
with such a small coefficient.
52
Table 9: Models 1 and 2 - Frequency of Terrorism, Overall, Algeria
53
Regarding the controls, only Model 1 showed any statistical significance.
Unexpectedly, gdp_change had a positive relationship with the frequency of terrorism. While
counter intuitive, the results were marginal with a coefficient of .458 with a 5% significance level.
Equally surprising, increases in population led to -2.526 with a 10% significance level. In contrast,
Increases in urbanization had expected results, leading to an increase in frequency with a
coefficient of 12.103 significant at the 5% level. This is nearly 3 times the average frequency of
attacks.
To summarize, in support of the theory, conciliation with leaders had the unintended
consequence of increasing the frequency of attacks with a 1 month lag. Additionally, in support of
the theory, violence against leaders decreased the frequency of attacks with a 1 month lag. In
opposition to the theory, violence against groups decreased the frequency of attacks with a 3
month lag. Model 1 and Model 2 suggest conciliation can have potentially negative repercussions
from a policy standpoint while violence can potentially reduce the frequency of terrorism.
Models 1 and 2 Severity
The results do not show statistical support for H1B, H2B, and H3B (See Table 10: Models
1 and 2: Effect of Gov. Responses on the Sev of Terrorism, Overall, Algeria). However, there was
support for H1D, H4B, and H4D with Model 1 and its 1 month lag. In support of H1D,
conciliation with leaders was significant at the 5% level with a coefficient of 52.840. Since,
Algeria averaged 29.470 casualties a month, this would represent nearly a 55% increase in
casualties. In support of H4D, violence against leaders had a coefficient of -20.167 with a 10%
significance level. Considering the average for monthly casualties, Algeria would experience a
decrease of approximately 46% in casualties following violence targeting terrorist leaders.
54
Model 2 with a 3 month lag had results in support of H4D; however, the real life implications
were minor. Violence against groups led to a decrease of -.352 in the number of casualties with a
significance level of 1%. While in support of the theory, in terms of lives the reduction would be
minor.
Regarding the controls, only Model 1 showed any statistical significance. Unexpectedly,
gdp_change had a positive relationship with the frequency of terrorism with a coefficient of 6.539
with a 5% significance level. Increases in population decreased casualties by -59.324 with a 10%
significance level. This is surprising since it was assumed increases in population would strain the
government's ability to maintain control and increase the size of the potential recruitment pool.
Less surprisingly, increases in urbanization resulted in an increase in the severity of attacks by
214.205 casualties. While expected since urbanization would assist terrorist mobility, an increase
of this degree is unexpected and is most likely an outlier at least in magnitude.
To summarize, in support of the theory, conciliation with leaders had the unintended
consequence of increasing the severity of attacks with a 1 month lag. In support of the theory,
violence against leaders decreased the severity of attacks with a 1 month lag. Additionally in
support of the theory, violence against groups decreased the severity of attacks with a 3 month
lag. Model 1 and Model 2 suggest conciliation can have potentially negative repercussions from a
policy standpoint while violence can potentially reduce the severity of terrorism.
55
56
Table 10: Models 1 and 2 - Severity of Terrorism, Overall, Algeria
Results by Hypotheses
Model 3
H1A: Conciliation with groups will increase the frequency of terrorist attacks during a
four-month period.
The results do not support H1A, acts of conciliation did not increase the frequency of
terrorism. Table 11 presents the effects of conciliation for Algeria overall as well as the effects
upon GIA, AQIM, and unknown perpetrators. For presentation purposes, the coefficients and
their significance levels were gathered from separate regressions and placed on individual charts.
The original regressions can be found in Appendix A: Regressions for Algeria.
Table 11: Effects of Conciliation on the Frequency of Terrorism, Algeria
Terrorist Org. Initial Month 1 Month Lag 2 Month Lag 3 Month Lag 0.758 -4.633 2.835 1.730
GIA (1.660)* (0.410) (0.210) (0.100) 0.616 -2.724 1.932 1.495
UNK (0.420) (3.60)*** (1.680)* (1.840)* Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
The effect of Conciliation on the overall frequency of attacks was significant at the 1%
level with a coefficient of -4.633 with a 1-month lag. Its negative relationship dissipates with a 2-
month lag. Following an act of conciliation, the following month saw a reduction in the
frequency of attacks well below the monthly average of 4.1970; however, the benefits disappear
57
58
after a 2 and 3 month lag with coefficients of 2.835 and 1.730. The positive relationships are
however statistically insignificant. Considering these results, while conciliation may reduce the
frequency of terrorism over the course of 4 months; the real life implications may be marginal.
Unexpectedly AQIM had a reduction of -0.819 with a 1-month lag at a 10% significance
level.24 This could be explained by the GSPC’s less radical ideology in the earlier years of the
investigation, at least relative to its later post-shift to AQIM.25 The GIA did see a slight increase
in the initial month, however this was only significant at the 10% level. As with the overall
frequency, the effect of conciliation on unknown perpetrators was significant at the 1% level
with a coefficient of -2.724. However, the frequency of unknown perpetrators later increase with
2 and 3 month lags. It appears as though, conciliation could reduce the overall frequency as well
as the frequency from unknown perpetrators in initial month. However, due to the increases with
2 and 3 month lags, the real life effects may be marginal. Considering these mixed results, H1A
remains unsupported.
H1B: Conciliation with groups will increase the severity of terrorist attacks during a four-month period.
As with H1A, the results show mixed results. Table 12 presents the effects of conciliation
for Algeria overall as well as the effects upon GIA, AQIM, and unknown perpetrators.
Conciliation decreased the severity of unknown perpetrators with a coefficient of -16.921 at the
1% significance level with a 1-month lag. This reduction would reduce severity from unknown
24 The regressions for all Algerian terrorism, AQIM, GIA, and UNK contained the same variable, concil_group_all because the manner in which the Algerian government offered conciliation, treating all terrorists equally regardless of affiliation. 25 The GSPC, as AQIM was originally known as, was created in reaction to the GIA’s attacks on civilians. As it became more affiliated with al-‐Qaeda, it too became more radicalized.
59
perpetrators well below its average of 14.083. As with the frequency of attacks, reductions with a
1-month lag were balanced by increases with a 2-month lag. Effects dissipated with a 3-month
lag. Conciliation increased the overall severity as well as the severity from unknown perpetrators
with coefficients of 28.773 and 11.153 with a 2-month lag. These effects were significant at the
5% level. Once again, it appears as though conciliation has beneficial effects that quickly
dissipate with the net affect canceling each other out over the course of a four-month period.
Due to the degree of increase at the overall level over the course of 4 months, the number of
casualties may be slightly higher. H1B finds weak support.
Table 12: Effects of Conciliation on the Severity of Terrorism, Algeria
Terrorist Org. Initial Month 1 Month Lag 2 Month Lag 3 Month Lag -8.356 -24.779 28.773 -4.974
GIA (1.700) (2.340)** (1.840) (0.200) 12.931 -8.427 -14.769 2.229
UNK (0.950) (0.780) (1.800) (0.300) Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
Controls
For the most part, the majority of controls were statistically insignificant. Ramadan
reduced the frequency of unknown perpetrators by -.925 with a 10% significance level. It also
reduced the severity of AQIM attacks by -13.372 casualties with a 5% significance level. This is
less then AQIM’s monthly average of 10.652. While this may seem unsurprising considering
most Algerians are Muslims, one would expect greater reductions in the overall levels.
An increase in urbanism was associated with a reduction in casualties from unknown
terrorists. This relationship was significant at the 5% level with a large coefficient of -113.1502.
This may be explained by an unexpected decrease in urbanism in the year 2000, one of the most
violent years in the 10 year period.
The severity of unknown perpetrators increased with increases in population. Pop_lag
had a coefficient of 25.326 with a 10% significance level. This is approximately 10 more then
the average for unknown perpetrators. This means that for every increase of 100,000 in
population size, the severity of unknown attacks would increase by 10. There probably is nothing
71
special about the relationship between unknown attacks and population size. It probably reflects
reflects the quality of open source data rather than an actual increased desire by terrorist groups
to hide their identities.
Unexpectedly, increases in the change of annual GDP were associated with increases in
the frequency and severity of GIA attacks. GDP_change increased both the frequency and
severity of GIA attacks at the 10% level with respective coefficients of .292 and 2.090. This is
most likely explained by increases in GIA attacks during improved economic conditions in 2002
and 2003.
.
72
CHAPTER FIVE: THE PHILIPPINES
Introduction
The following will assess governmental responses in the Philippines. It will use a similar
format as the preceding chapter. The first section of the chapter provides a brief background to
the conflict. The results of the analysis will then be presented. Models 1 and 2 will be discussed
at the same time while focusing on areas of statistical significance. Since Model 3 is far more
comprehensive, it will be discussed in detail while addressing each hypothesis individually.
All the models include the same dependent variables, frequency and severity as well as
the same governmental responses, and controls. Models 1 and 2 only investigate overall levels of
frequency and severity. Model 1 includes a 1 month lag. Model 2 includes a 3 month lag. In
contrast, Model 3 also investigates effects upon individual groups. Model 3 includes multiple
lags to include the initial month, 1, 2, and 3 months lags.
Given the ten-year scope of the investigation, there was not enough observations to assess
responses on an annual basis. However, potential long term effects, particularly those of
conciliation and violence, are discussed in the final chapter. Discussion points, final comments,
gaps in the theory, and potential for future research will be discussed in the final chapter.
73
Background
In the 16th century, the Philippines become a Spanish colony. The centuries of Spanish
colonialism had obvious political effects, deterring organic political development, but it also had
cultural ramifications, fragmenting society (BBC, 2012; Country Profile: Philippines). Over 82
percent of Filipinos are Roman Catholic. There is a 5 percent Muslim minority concentrated
mainly on the southwestern portion of Mindanao Island and the southwestern archipelagos (CIA
Factbook).
After the Spanish American War of 1898, the Philippines entered a period of U.S.
military rule that continued until 1935. Under the Commonwealth, the Philippines experienced
self-rule for the first time in centuries. Following World War II, in 1946, like so many other
colonies, the Philippines became fully independent (BBC, 2012; Rogers 2004).
Unfortunately, independence did not mean stability or political freedom. Angered by a
lack of land reform, the Huks, former members of the People’s Anti-Japanese Army, a wartime
militia with a communist ideology, launched an insurgency that peaked between 1949 and 1951.
A lack of training and arms sapped their initial momentum and within a few years, the entire
movement dissolved (Dolan, 1991).
Peace however did not last long. After being elected President in 1965, Ferdinand Marcos
faced renewed opposition from communists in the north and Moro separatists in the south during
the late 1960s. Threatened by these dual threats, in 1972, Marcos suspended Parliament and
placed the entire country under martial law. Expanded presidential powers were established under
a new constitution the following year (Library of Congress, 2006). Martial law was finally lifted
in 1981. This however did not prevent the assassination of Marcos’s main opponent, Benigno
74
Aquino in 1983 (BBC, 2012). In 1986, the Philippines took a step towards democracy when
mass demonstrations, diminished military support, and pressure from the Catholic church forced
Marco’s resignation and flight to the United States. Ironically, he was replaced by Corazon
Aquino, the widow of his former rival (BBC, 2012; Library of Congress, 2006).
Elections did not guarantee a sense of legitimacy or stability. In 2000, President Estrada
narrowly survived impeachment charges of corruption. Once again mass demonstrations took to
the streets and forced the President to leave office in 2001. His vice-president, Gloria Arroyo,
took over the presidency and won reelection in 2004 despite allegations of corruption. After
gaining independence, corruption and cronyism have been par for the course. Since 1965, the
Philippines have had five presidents; out of those, two have resigned, two have been arrested,
and one has physically fled the country (BBC, 2012; Library of Congress, 2006; Dolan, 1991;
Hedman, 2006).
Beyond corruption, combating insurgency and terrorism face challenges inherent in the
cultural, geographic, and political characteristics of the Philippines. Geographically, the
Philippines is comprised of over 7,000 islands (CIA Factbook). While island geography is
supposed to favor the counterinsurgent, this may not hold true when there are thousands of
islands or the indigenous population has a long maritime history (Galula 1969). The traditional
term for village or community, barangay, is actually Malay for boat and it represents a long
history of fishing and nautical navigation. Galula also assumes the counterinsurgent has an
adequate navy to patrol its shores. As of 2006, the Philippines had less than 70 ships with only
half of those serviceable (Library of Congress, 2006). With 36,289 kilometers of coastline, that
leaves 1451 kilometers per operational vessel to patrol (CIA Factbook).
75
Culturally, the Philippines remain ethnically and religiously divided. While the term
“Moro” is derived from the Spanish term for Moors, the Moros are also ethnicially distinct.
(Library of Congress, 2006). Moro society itself is deeply divided along ethnic and tribal lines.
Moro communities also have the highest levels of poverty and unemployment in the Philippines
(Jones & Morales 2012). For instance, the Basilan and Jolo provinces with Moro majorities are
two of the worst in terms of development. Land ownership is also problematic. In Basilan,
Muslims comprise 71 % of the population yet only own 25 % of the land (Donnelly 2004).
Changes in traditional power structures have also left power vacuums in Moro society.
Historically, Moro communities were led by a datu or chieftain whose power was based upon
kinship ties and the numbers of supporters. In order to increase these numbers, villages often
raided their neighbors (Dolan, 1991). Ironically, centralized power and the delegitimization of
violence have eroded the traditional power structures within Moro society. Village and tribal
leaders are no longer able to use violence to consolidate their power. The current situation
particularly in the peripheral islands suffers from weak local leaders and an absent government.
As a result, far too often even when leaders seek peace, they have found it almost impossible to
control their supporters (Collier 2006).
While there have been numerous terrorists groups in the Philippines, due to the above
challenges, resistance from the Moro minority has been problematic. In 1972, Nur Misuari
created the Moro National Liberation Front (MNLF) in order to achieve an independent state for
the Moro people. In 1976, MNLF agreed to a referendum to determine the future of the southern
islands with Muslim minorities; however, the effort failed due to the Christian majority (Jones &
Morales 2012; Memorial Institute for the Prevention of Terrorism, 2006). Eventually, a truce was
76
signed, and in 1996, the government created the Autonomous Region of Muslim Mindanao with
Misuari as its leader.26 While these negotiation failed to end terrorism, they did successfully split
the Moro opposition. In 1978, after unsuccessfully challenging Misuari for MNLF leadership,
Hashim Salamat created the Moro Islamic Liberation Front, with greater emphasis on Islam
(Memorial Institute for the Prevention of Terrorism, 2006; Rogers 2004).27
Following military operations and the mass surrender of nearly two thousand members,
the MILF signed peace agreements with the government in 2001 and 2003 (Banos, 2006).
However, noncompliant members continued to conduct terrorism and associate with members of
both ASG and JI (Hedman 2006). While, the MILF has officially renounced terrorism and
actually conducted joint operations against ASG and JI; terrorism has continued to kill hundreds
of individuals every year (Hedman 2006).
In addition to the MILF, the Philippines also suffers from transnational religious
terrorism. The ASG was formed by Abdurajak Janjalani, who was suspected of ties with Al-
Qaeda, in order to create pan-Islamic state from Mindanao to the southern portion of Thailand
(Memorial Institute for the Prevention of Terrorism, 2006; Rogers 2004). It splintered from the
MNLF after the creation of the autonomous region. While it still verbalizes Islamic ideology, it
has devolved into a criminal organization dependent upon ransom and extortion (James &
Morales 2012; Rogers, 2004). In 1998, Philippine security forces killed Janjalani. With the death
of its founder, the ASG has become increasingly decentralized. This decentralization potentially
26 After connecting Misuari to a 2001 terrorist attack, the government placed Misuari under house arrest (“Moro National Liberation Front”). 27 The MILF was originally called the “new MNLF.” In 1984, it officially changed its name to MILF (Trager & Zagorchev 2005/2006).
77
limits the impact of eliminating ASG leaders (Memorial Institute for the Prevention of
Terrorism, 2006).
The last group, Jemaah Islamiya (JI), was created in 1993 as an Islamic terrorist group. It
gained notoriety after the 2002 Bali bombing. Although it primarily operates in Malaysia, it also
operates in the Philippines and has ties with renegade members of MILF and ASG. In 2003, one
of its top leaders, Hambali, was captured by Philippine security forces (Memorial Institute for the
Prevention of Terrorism, 2006; Jones & Morales, 2012).
In recent years, the threat of communist terrorism, particularly from the New People’s
Army (NPA) has eclipsed Islamic terrorism. Formed in 1969, the NPA is the armed wing of the
Communist Party of the Philippines, Marxist-Leninist (CPP-ML). As an insurgency, it reached
its height between 1969 and 1976 while receiving support from China. After losing its benefactor
and a series of military defeats, the NPA has transformed into a terrorist organization that is
increasingly associated with criminal activity. In order to weaken the government, the NPA has
sought to dissuade foreign investments by attacking foreign companies. It also has a history of
assassinating political opponents (Memorial Institute for the Prevention of Terrorism, 2006).
Negotiations between the government and communist factions took place in Norway in 2004.
However, they quickly broke down due to the NPA’s insistence that the United States remove
the Communist Party of the Philippines (CPP) and NPA from its list of foreign terrorists
(Hedman, 2006). In 2007, the government offered an amnesty program specifically to the
communists without preconditions beyond individuals forgoing future terrorist activities.
78
Descriptive Statistics
Table 23 provides the descriptive statistics for both the independent and dependent
variables. Overall, the Philippines averaged 3.886 attacks with an average of 18.773 casualties a
month reaching the highest levels of violence in 2003 during negiotations with the MILF (See
Figure 3: Overall Frequency and Severity of Terrorism by, the Philippines). Overall, the MILF
was the deadliest, averaging 1.197 attacks and 9.788 casualties a month, three more than the
overall average. In contrast, the NPA averaged 1.114 attacks with 2.644 casualties a month. The
casualty rate from NPA attacks is almost a quarter of that from MILF attacks. The second
deadliest group, in terms of casualty rates per attacks, ASG, averaged .614 attacks and 6.932
casualties a month. JI averaged only .091 attacks resulting in 1.780 casualties a month.
Figure 3: Overall Frequency and Severity of Terrorism by, the Philippines
In terms of governmental violence, the government killed or arrested an average 25.750
terrorists a month. In comparison, the government used violence against the MILF an average 11
times a month. This is nearly double the use of violence against the NPA. As was the case with
Algeria, this probably reflects the government targeting the most lethal group, as opposed to
potential bias against Islam (See Table 23: Descriptive Statistics for Governmental Responses,
the Philippines).
79
80
Table 23: Descriptive Statistics for Governmental Responses, the Philippines
vio_group_unk 6.258 9.521 0.000 45.000 Note: The above figures only include data from 2000-2010. Data was collected to allow for lags between September 1999 and December 1999. Additionally, denial and legal restriction are not broken down by group since Algerian applied these responses equally to all groups. (n = 132).
81
Results
The results are mixed requiring an understanding of the particular circumstances in the
Philippines during the period of investigation. The results for the overall levels of frequency and
severity at the national level as well as for specific terrorist groups were achieved using separate
regressions. Models 1 and 2 will be discussed together while focusing on areas of statistical
significance. Frequency will be discussed first followed by severity. The regressions can also be
found in Appendix B (Model 1 and 2). Model 3 will be discussed by itself while addressing
individual hypotheses. In order to facilitate the review of multiple lags, specific results were
taken from separate regressions to form the tables. The original regressions for Algeria can be
found in Appendix D. All regressions used robust standard errors.
Models 1 and 2 Frequency
The results do not show statistical support for H1A, H1C, H2A, H4A, and H4C (See
Table 24: Models 1 and 2: Effect of Gov. Responses on the Freq of Terrorism, Overall, the
Philippines). However, there was support for H3A with Model 1 and its 1 month lag. In support
of H3A, legal restriction was significant at the 10% level with a coefficient of -2.335. Since the
Philippines averaged 3.886 attacks a month, legal restriction would decrease the frequency of
attacks by 60% a month. Relative to Algeria, legal restriction in the Philippines was more
successful most likely due to the increased detention powers provided to police forces.
In contrast, Model 2 with a 3 month lag had opposite results. With the additional lag, legal
restriction actually led to an increase in the frequency of attacks. With a coefficient of 2.791
significant at the 1% level, legal restriction actually increased the frequency of attacks. This
contradiction could be explained by terrorists adopting to the new tactical situation created by
legal restriction.
Regarding the controls, only ramadan showed any statistical significance in both Models
1 and 2. Since Islamic terrorists conducted a large percentage of the terrorist attacks in the
Philippines, Ramadan could be expected to decrease the frequency of attacks. In Model 1,
ramadan had a negative coefficient of -1.936 and a significance level of 5%. Model 2 showed
similar results with a negative coefficient of -1.689 and a significance level of 5%.
82
Table 24: Models 1 and 2 - Frequency of Terrorism, Overall, the Philippines
83
Models 1 and 2 Severity
84
The results do not show statistical support for H1B, H1D, H2B, H4B, and H4D (See Table
25: Models 1 and 2: Effect of Gov. Responses on the Freq of Terrorism, Overall, the Philippines).
Only the findings regarding legal restriction had any statistical significance. In Model 1 with a 1
month lag, in contrast to H2B, legal restriction decreased the severity of terrorism by -14.433,
significant at the 5% level. While in Model 2 with a 3 month lag, legal restriction had the opposite
effect increasing the severity of terrorism by 24.364, significant at the 5% level. While supportive
of H2B, it is problematic since the frequency also increased. It appears after 3 months, legal
restriction led to both an increase in the frequency and the severity of terrorism. None of the
controls were statistically significant.
Table 25: Models 1 and 2 - Severity of Terrorism, Overall, the Philippines
85
86
H1A: Conciliation with groups will increase the frequency of terrorist attacks during a four-month period.
The results show weak support for H1C. Table 26 presents the effects of conciliation
with groups upon the overall frequency of attacks in the Philippines, as well as the impacts upon
the frequency of attacks conducted by the ASG, JI, MILF, NPA, and unknown perpetrators.28
Regressions for the Philippines in their entirety can be found in Appendix B: Regressions for the
Philippines.
Conciliation had a positive relationship with the overall frequency of terrorism in the
initial month with a significance level of 5% and a coefficient of 3.068. If this increase is added
to the average number of overall attacks, the Philippines would experience approximately 7
attacks in a month, nearly twice as much as the monthly average. The frequency of unknown
perpetrators also increased with a coefficient of .734 with a 10% significance level. These effects
however dissipated in significance levels with any lag.
28 The Philippine government did not negotiate with ASG or JI. However, it reached several deals with the MILF. One of the major goals of these agreements was to increase cooperation between the government and the MILF to reduce ASG and JI attacks. For this reason, agreements with the MILF were also included in the regressions for ASG and JI.
Results by Hypotheses
Model 3
87
Table 26: Effects of Conciliation on the Frequency of Terrorism, the Philippines
Terrorist Org. Initial Month 1 Month Lag 2 Month Lag 3 Month Lag 3.068 1.069 1.002 0.331
Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
89
H1C: Conciliation with leaders will result in an increase in the frequency of terrorist attacks during a four-month period to a greater extent then conciliation with groups.
The results do not support H1C. Table 28 presents the effects of conciliation with leaders
upon the overall frequency of attacks in the Philippines, as well as the impacts upon the
frequency of attacks conducted by the ASG, JI, MILF, NPA, and unknown perpetrators.
Contrary to the theory as well as the Algerian results, conciliation with leaders resulted in
reductions in the overall frequency as well as the frequency of ASG and NPA attacks. The
frequency of overall attacks was below its average of 3.886 with a 2-month lag and significant at
the 5% level. While the frequency of NPA attacks was also below it’s average of 1.14 with a 3-
month lag and 5% significance level. This is particularly surprising since the effects are even
greater than conciliation with groups however they are delayed until the 2 and 3 month lags.
Perhaps Philippine terrorist leaders were more willing than their followers to negotiate, and it
took several months before they could gain control over their followers.
90
Table 28: Effects of Conciliation towards leaders on the Frequency of Terrorism, the Philippines
Terrorist Org. Initial Month 1 Month Lag 2 Month Lag 3 Month Lag 0.946 3.246 -4.203 1.516
UNK (0.490) (1.550) (0.830) (1.580) Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
H1D: Conciliation with leaders will increase in the severity of terrorist attacks during a four-month period to a greater extent then conciliation with groups.
The results regarding H4C are mixed. Table 29 presents the effects of conciliation with
leaders upon the overall severity of attacks in the Philippines, as well as the impacts upon the
severity of attacks conducted by the ASG, JI, MILF, NPA, and unknown perpetrators. Initially,
conciliation with leaders led to increases in the severity of attacks for both the Philippines overall
and unknown perpetrators at the 5% and 1% levels. This however was countered by decreases in
overall levels and the severity of NPA attacks at the 5% and 1% levels at 2 and 3 month lags,
respectively.
91
Table 29: Effects of Conciliation with leaders on the Severity of Terrorism, the Philippines
Terrorist Org. Initial Month 1 Month Lag 2 Month Lag 3 Month Lag 11.407 16.810 -26.326 -3.793
Note: The unit of analysis was on a monthly basis however for the purposes to presentation the data was aggregated to the annual level.
Concluding Remarks
Several takeaways are important to mention. Offers of amnesty may be more effective
when applied to insurgents then terrorists. In Algeria it took three years before the overall levels
of frequency and severity were significantly reduced. However, it did successfully drain
thousands of supporters from an insurgency that at one time poised an existential threat to the
government. Similarly, with major agreements reached with the MILF including cooperation
against other extremist groups, the Philippines successfully lowered terrorism levels. In the long
109
term, the 2003 agreement with the MILF appears to have led to significant reductions in both
frequency and severity.
Offering benefits to groups while attributing costs to leaders appears to have the greatest
benefits. Unfortunately, the investigation failed to assess government actions towards the support
population. This was the result of two factors. First, the coding process was unable to identify
concessions to the support population. While there were instances such as payments to victims of
government action, it was impossible to code them on a monthly basis. This remains the most
serious limitation of the method and the investigation as whole. Second, it appears neither the
Algerian or Philippine governments incorporated concessions to the support populations as
major parts of their counterterrorist policies. Regardless of the particular reason, the interaction
between the government and the support population was not adequately assessed. This could be
addressed by either adjusting the coding process or by including additional cases.
While the project investigated effects upon specific groups, this process was severely
hampered by the nature of the data. Due to the presence of an extreme amount of unknown
perpetrators in the GTD, any conclusions regarding specific terrorist groups should be
considered with hesitation. This by no means is a slight to the GTD; it merely reflects the data is
only as good as the open sources that originally reported the information. Unfortunately due to
the large number of unknown perpetrators, assessing the success or failure of government policy
towards individual groups is problematic. With these issues acknowledged, the results at least
suggest internal dynamics and the context of governmental action are more important than
ideological factors. There did not seem to be any pattern of results among Islamic groups. This
casts doubt upon Miller’s assertion that governmental responses need to be tailored to a group’s
ideological beliefs (2007).
110
The two newly developed databases allow several topics for future research. First, it
would be beneficial to expand the scope of the investigation. This is particularly true in regards
to Algeria. By expanding the scope to twenty years, it would be interesting to see how the pattern
of violence shifted from the 1990s during periods of insurgency to later when the conflict was
dominated by terrorism. It would also be interesting to isolate certain time periods. For instance,
it appears conciliation had different effects whether one compares the amnesty of 2000 to the
amnesty of 2006. After regressing the same model with data between 2000 and 2005,
conciliation had the effect of reducing the frequency of overall terrorism by nearly 5 attacks with
a 1 month lag and a significance level of 1%. When comparing a similar regression between
2006-2010, conciliation only reduced the frequency of overall terrorism by 2 attacks with a
significance level of less then 10%.30 This may be the effect of negative reputation. Since the
government reportedly did not honor its commitment to expand political participation, perhaps
terrorists were less likely to participate in future amnesties. Or it could reflect increased
radicalization or isolation from the very society, the terrorists wish to change, as was suggested
by Crenshaw (1981).
30 This is not formally presented as a regression but merely serves as a point of discussion for future research.
Methodologically, there are three main areas to improve the research design. First, by
including the initial month in Model 3, there may be confusion in what is actually happening first,
the independent variable or the dependent variable. For instance, with denial, increases in
terrorism may have occurred first followed by the government's response, roadblocks. Even if
they successfully decreased terrorism, it would appear as the opposite. Second, conciliation with
groups could be coded as a simple dummy variable rather than a categorical variable. Third, in
111
This research assessed four common governmental responses: conciliation, denial, legal
restriction, and violence. Conciliation appears to have mixed results. In general, the results show
that conciliation may led to increases in terrorism in the short term while suggesting potential
reductions in the long term. Concessions such as amnesties without political concessions do not
address root grievances and therefore their effects may be limited. Denial and legal restriction
often led to increases in terrorism, merely forcing terrorists to shift targets. While the effects of
violence often depended upon whether it was applied to groups or their leaderships.
Unfortunately, the most successful policies, whether those of conciliation or violence, often
drove defectors to more radical, lethal groups. At this point, governments must be willing and
capable to engage these groups violently, concentrating their efforts on terrorist leaders.
Because the databases cover some responses in even more detail than what was covered in
this study, a series of investigations could explore one response at a time in greater depth. For
instance, the effects of arrests and kills could be separately investigated. Similarly, it would be
interesting to see if arresting leaders is more beneficial than killing them, due to increased
intelligence gains following interrogation. Research could also investigate whether conciliation
causes terrorist infighting and does that infighting actually affect the frequency and severity of
terrorism.
order to avoid too few events per variable as was often the case with conciliation with leaders,
legal restriction, and violence with leaders. In order to avoid variables with too few events, future
research could combine conciliation with groups with conciliation with leaders and violence
against groups with violence against leaders. Unfortunately, legal restriction unless defined
differently, would have to be dropped.
112
APPENDIX A ALGERIA - MODELS 1 & 2
Frequency
Table 1: Models 1 and 2 - Frequency of Terrorism, Overall, Algeria
113
Severity
Table 2: Models 1 and 2 - Severity of Terrorism, Overall, Algeria
114
115
APPENDIX BTHE PHILIPPINES - MODELS 1 & 2
Frequency
Table 1: Models 1 and 2 - Frequency of Terrorism, Overall, the Philippines
116
Severity
Table 2: Models 1 and 2 - Severity of Terrorism, Overall, the Philippines
117
118
APPENDIX C ALGERIA - MODEL 331
31 The Appendix is ordered by frequency and then severity. Regressions for the overall levels are presented first, followed by individual organizations in alphabetical order.
119
Frequency
Table 1: Effect of Gov. Responses on the Freq of Terrorism, Overall, Algeria
R-Squared Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
127
APPENDIX D THE PHILIPPINES - MODEL 332
32 The Appendix is ordered by frequency and then severity. Regressions for the overall levels are presented first, followed by individual organizations in alphabetical order.
128
Frequency
Table 1: Effect of Gov. Responses on the Frequency of Terrorism, Overall, the Philippines
R-Squared Note: Absolute value of t-statistic in parentheses. *p ≤ .10; **p ≤ .05; ***p ≤ .01 (two-tailed tests). n = 132.
140
REFERENCES
Adams, Kenneth. (2003). The effectiveness of juvenile curfews at crime prevention. The Annals of the American Academy of Political and Social Science, 587, 136-159.
AFP. (2010, February 25). Algerian police chief shot dead during row. Agence France Presse.
AFP. (2004, August 25). Five Algerian soldiers, two police killed in ambush. Agence France Presse.
Al Jazeera. (2011 January 5). Tunisian protestor dies of burns. Al Jazeera. Retrieved from http://www.aljazeera.com/news/africa/2011/01/201115101926215588.html.
Amos II, John and Russel H.S. Stolfi. (1982). Controlling international terrorism: alternatives palatable and unpalatable. International Terrorism, 69-83.
Bar, Shmuel. (2008). Deterring terrorism: what Israel has learned.” Hoover Institution Stanford University. Retrieved from http://www.hoover.org/publications/policyreview/article/5674
BBC. (2012, September 4). Profile: Algeria. BBC News. Retrieved from http://www.bbc.co.uk/news/world-africa-14118854.
BBC. (2001, December 28, 2001). Hit US economy, urges Bin Laden. BBC News, Retrieved from http://news.bbc.co.uk/2/hi/south_asia/1731112.stm.
BBC. (2009, April). Landslide win for Algeria. BBC News. Retrieved from http://news.bbc.co.uk/2/hi/africa/7993671.stm.
BBC. (2003, March 13). Palestinians get Saddam Funds. BBC News. Retrieved from http://news.bbc.co.uk/2/hi/middle_east/2846365.stm.
BBC. (2012, December 12). Philippines Profile. BBC News. Retrieved from http://www.bbc.co.uk/news/world-asia-15521300.
BBC. (2003, April 29). US pulls out of Saudi Arabia. BBC News. Retrieved from http://news.bbc.co.uk/2/hi/middle_east/2984547.stm.
Berman, Eli, Jacob N. Shapiro, and Joseph H. Felter. (2011). Can hearts and minds be bought? the economics of counterinsurgency in Iraq. Journal of Political Economy, 119 (4), 766-819.
141
Betts, Richard K. (2002) The soft underbelly of American primacy: tactical advantage of terror." Political Science Quarterly, 117 (1), 19-36.
Bin Laden, Osama. (2002). Full text: Bin Laden's letter to America.” The Guardian. Retrieved from http://www.guardian.co.uk/world/2002/nov/24/theobserver.
Bjørgo, Tore. (2011). Dreams and disillusionment: engagement in and disengagement from militant extremist groups. Crime Law Social Change, 55, 277–285.
Buenos de Mesquita, Bruce, Alastair Smith, Randolph Siverson, and James Morrow (2004). The logic of political survial. Cambridge, MA: MIT Press, 38.
Bueno de Mesquita, Ethan. (2005). Conciliation, counterterrorism, and patterns of terrorist violence.” International Organization, 59 (1), 145-176.
Bueno de Mesquita, Ethan. (2005). The Quality of Terror. American Journal of Political Science, 49, (3), 515–530.
Bueno de Mesquita, Ethan and Eric S. Dickson. (2007). The propaganda of the deed: terrorism, counterterrorism, and mobilization. American Journal of Political Science, 51 (2), 364-381.
Caplan, Bryan. (2006). Terrorism: the relevance of the rational choice model. Public Choice, 128 (1/2), 91-107.
Carter, Jeremy G. and David L. Carter. “Law Enforcement Intelligence: Implications for Self-radicalized Terrorism.” Police Practice and Research: An International Journal Vol. 13, no. 2 (2012): 138-154.
Cederman, Lars-Erik and Luc Girardini. (2007). Beyond fractionalization: mapping ethnicity onto nationalist insurgencies.” The American Political Science Review, 101, (1), 173-185.
Central Intelligence Agency. (2013). The World Factbook: Philippines. Retrieved from https://www.cia.gov/library/publications/the-world-factbook/.
Collier, Kit. (2006). Terrorism: evolving regional alliances and state failure in Mindanao. Southeast Asian Affairs, 26-38.
Collier, Paul and A. Hoeffler. (2004). Greed and grievance in civil war.” Oxford Economic Papers, 56, (4), 563-595.
Congressional Record. (2001). US Patriot Act: Public Law 107-56. Washington, DC.
142
Crenshaw, Martha. (1981) “The Causes of Terrorism.” Comparative Politics, 13 (4), 379-399.
Cronin, Audrey Kurth. (2006). How Al-Qaida ends: the decline and demise of terrorist groups. International Security, 31, (1), 7-48.
Davis, Paul. (2010). Simple models to explore deterrence and more general influence in the war with al-Qaeda. Santa Monica, CA: RAND Corporation.
Davis, P. K., and Brian Michael Jenkins. (2002). Deterrence & influence in counterterrorism: a component in the war on Al Qaeda. Santa Monica, CA: Rand Corporation.
Dolan, Ronald E. (1991). Philippines: A Country Study. Washington: GPO for the Library of Congress.
Donnelly, Charles. (2004). Terrorism in the southern Philippines: contextualizing the Abu Sayyaf Group as an Islamist secessionist organization. 15th Biennial Conference of the Asian Studies Association of Australia. Canberra, Australia.
Dugan, Laura and Erica Chenoweth. (2012). “Moving beyond deterrence: the effectiveness of raising the expected utility of abstaining from terrorism in Israel.” American Sociological Review, 77, 597-624.
Dugan, Laura and Erica Chenoweth. (2012). Government actions in a terror environment – Israel (GATE-Israel). National Consortium for the Study of Terrorism and Responses to Terrorism. Retrieved from http://hdl.handle.net/1902.1/18903UNF:5:jPRWAw/AaA2-llTkaFQJfPg== National Consortium for the Study of Terrorism and Responses to Terrorism.
Enders, Walter and Todd Sandler. (1993). The effectiveness of antiterrorism policies: a vector- autoregression-intervention analysis.” The American Political Science Review, 87 (4), 829-844.
England, Vaudine. (2008). Is Philippine peace process dead? BBC News. Retrieved from http://news.bbc.co.uk/2/hi/asia-pacific/7599853.stm.
Esterbrook, John. (2009, February 11). Salaries for suicide bombers. CBS News, Retrieved from http://www.cbsnews.com/2100-202_162-505316.html.
Fearon, James. (1997). Signaling foreign policy interests: tying hands versus sinking costs. The Journal of Conflict Resolution, 41 (1), 68-90.
Fearon, James. (1994). Signaling versus the balance of power and interests: an empirical test of a crisis bargaining. The Journal of Conflict Resolution, 38, (2), 236-269.
143
Fearon, James, Kimulikasara and David D. Laitin. (2007). Ethnic minority rule and civil war onset. The American Political Science Review, 101 (1), 187-193.
Fearon, James and David D. Laitin. (2003). Ethnicity, insurgency, and Civil War.” American Political Science Review, 97 (1), 75-90.
Fisher, Uri. (2007). Deterrence, terrorism, and american values." Homeland Security Affairs, III (1) no. 1, 1-17.
Galula, David. (1964) Counter-insurgency warfare: theory and practice. New York, NY: Frederick A. Praeger.
Ganor, Boaz. (2011). An intifada in europe? a comparative analysis of radicalization processes among Palestinians in the West Bank and Gaza versus muslim immigrants in Europe. Studies in Conflict & Terrorism, 34, (8), 587-599.
Gordon, Philip H. (2007). Can the war on terror be won? how to fight the right war. Foreign Affairs, 86, (6), 53-66.
Gurr, Ted Robert. Terrorism in democracies: its social and political bases.” In Origins of Terrorism: Psychologies, Ideologies, Theologies, States of Mind. Washington D.C: Woodrow Wilson Center Press, 86. Edited by Walter Reich.
Hedman, Eva-Lotta. (2006). The Philippines in 2005: old dynamics, new conjuncture. Asian Survey, 46 (1), 187-193.
Heymann, Philip B. (2001-2002). “Dealing with Terrorism: An Overview.” International Security, 26 (3), 24-38.
Hippel, Karin von. (2008). A counter radicalization strategy for a new U.S. administration. The Annals of the American Academy of Political and Social Sciences, 618, 182-196.
Home Office: UK Border Agency. (2013 January 17). Country of origin information (COI) report: Algeria. Home Office, 12-18, 180-185.
Horgan, John. (2008). From profiles to pathways and roots to routes: perspectives from
144
psychology on radicalization into terrorism. The Annals of the American Academy of Political and Social Sciences, 618, 80-94.
Horgan, John, and Kurt Braddock. (2010). Rehabilitating the terrorists?: challenges in assessing the effectiveness of de-radicalization programs.” Terrorism and Political Violence, 22, (2) 267-291.
Humphreys, Macartan, and Jeremy M. Weinstein. (2006). Handling and manhandling civilians in civil war.” The American Political Science Review, 100, (3), 429-447.
Huth, Paul and Bruce Russett. (1984). What makes deterrence work? cases from 1900 – 1980.” World Politics, 36, (4), 496-526.
International Crisis Group. (2001, July 9). The civil concord: a peace initiative wasted. ICG Africa Report (31).
Jones, Clarke R. Jones, and Resurrecion S. Morales. (2012): Integration versus segregation: a preliminary examination of Philippine correctional facilities for de-Radicalization. Studies in Conflict & Terrorism, 35:3, 211-228.
Jones, Seth G., and Martin C. Libicki. (2008). How terrorist groups end: lessons for countering al Qaida. Santa Monica, CA: Rand Corporation.
Kalyvas, Stathis. (2009). The logic of violence in civil war. New York, NY: Cambridge University Press, 2009: 111-145.
Kalyvas, Stathis N. (2004). The paradox of terrorism in civil war.” The Journal of Ethics, 8 (1), 97-138.
Kavanagh, Jennifer. (2011). Selection, availability, and opportunity: the conditional effect of poverty. Journal of Conflict Resolution, 55, 106-132.
Kennedy, Jonathon. (2012). “Countermeasures against Extremism and Terrorism (CoMET) Database: Data Status and Final Findings.” Study of Terrorism and Responses to Terrorism (START). College Park, MD: START.
Kennedy, Jonathan W., and Dominic Lisanti, and Brooke Fisher Liu. (2012). Countermeasures against extremism and terrorism (CoMET). Database: Data Status and Final Findings.” Study of Terrorism and Responses to Terrorism (START). College Park, MD: START.
Kershner, Isabel, and Fares Akram. (2012, November 14). Ferocious Israeli assault on Gaza kills a leader of Hamas.” New York Times. Retrieved from http://www.nytimes.com/2012/11/15 - /world/middleeast/israeli-strike-in-gaza-kills-the-military-leader-of-hamas.html?pagewanted=all.
145
Kleinmann, S.M. (2012). Radicalization of homegrown Sunni militants in the United States: comparing converts and non-converts.” Studies in Conflict & Terrorism, 35 (4), 278-297.
Kroenig, Matthew and Barry Pavel. (2012). How to deter terrorism. The Washington Quarterly, 35 (2), 21-36.
Knopf, Jeffrey W. (2010). The fourth wave in deterrence research. Contemporary Security Policy, 31 (1), 1-33.
Krueger, Alan B., and Jitka Maleckova. (2003). Education, poverty and terrorism: is there a causal connection? Journal of Economic Perspectives, 17 (4), 119-144.
Kydd, Andrew H., and Barbara F. Walter (2006). "The Strategies of Terrorism." International Security, 31 (1), 49-79.
LaFree, Gary, and Laura Dugan. (2009). Research on terrorism and countering terrorism. Crime and Justice, 38 (1), 413-477.
Lawrence, T.E. (2008). Science of guerrilla warfare. In Strategic Studies. London: Rutledge, 2008, 251.
Lee, Dwight R. (1998). Free riding and paid riding in the fight against terrorism.” The American Economic Review, 78, (2), 22-26.
Library of Congress. (2006). Country Report: Philippines. Washington: GPO for the Library of Congress.
Loidolt, Bryce (2011): Managing the global and local: the dual agendas of Al Qaeda in the Arabian Peninsula. Studies in Conflict & Terrorism, 34:2, 102-123.
Lyall, Jason. (2009) Does indiscriminate violence incite insurgent attacks?: evidence from chechnya. Journal of Conflict Resolution, 53, 331-362.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Abu Sayyaf Group. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Al Qaeda. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Al Qaeda in the Islamic
146
Maghreb. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Armed Islamic Group. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Islamic Salvation Front. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Jemaah Islamiyah. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Moro Islamic liberation front. Terrorist Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). Moro national liberation front. Terrorist Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Memorial Institute for the Prevention of Terrorism (MIPT). (2008). New People’s Army. Organization Profile: Study of Terrorism and Responses to Terrorism. Retrieved from http://www.start.umd.edu/start/data_collections/tops/.
Mearsheimer, John (1995). Conventional Deterrence. Ithaca and London: Cornell University Press, 23-24.
Miller, Gregory. (2007). Confronting terrorisms: group motivation and successful state policies. Terrorism and Political Violence, 19, 331–350.
Mingst, Karen (2003). Essentials of International Relations. New York City: Norton & Company.
Morgan, Patrick Deterrence Now. (2003). New York City: Cambridge University Press.
Mueller, John and Mark G. Stewart. (2011). Terror, security, and money: balancing the risks, benefits, and costs of homeland security. Panel at Annual Convention of the Midwest Political Science Association. In "Terror and the Economy: Which Insitutions Help Mitigate the Damage?".
147
National Consortium for the Study of Terrorism and Responses to Terrorism (START). (2012). Global Terrorism Database [Data file]. Retrieved from http://www.start.umd.edu/gtd.
Oberschall, Anthony. (2004). Explaining terrorism: the contribution of collective action theory. Sociological Theory, 22, (1), 26-37.
Pape, Robert A. (2003). The stragetic logic of suicide terrorism. American Political Science Review, 97 (3), 1-19.
Perliger, Arie. (2012). How democracies respond to terrorism: regime characteristics, symbolic power and counterterrorism.” Security Studies, 21 (3), 490-528.
Porch, Douglas. (2012). The hunt for martin caballero. Journal of Strategic Studies, 35 (2), 243- 270.
Porter, L.E., and M.R. Kebbell. (2011). Radicalization in Australia: examining Australia's convicted terrorists. Psychiatry, Psychology and Law, 18 (2), 212-231.
Posen, Barry. (2001/2002). The struggle against terrorism: grand strategy, strategy, and tactics.” International Security, 26 (3), 39-55.
Post, J., & Sheffer, Gabriel (2007). The risk of radicalization and terrorism in U.S. Muslim communities.” Brown Journal of World Affairs, 101-112.
Renard, Thomas. (2008). AQIM’s offensive reveals shift from insurgency to terrorist tactics in Algeria. Terrorism Monitor, 6 (18).Retrieved from http://www.jamestown.org/programs/gta/single/?tx_ttnews%5Btt_news%5D=5171&tx_ttnews%5BbackPid%5D=167&no_cache=1.
Reuters. (2011, February 24). Algeria lifts 19-year-old state of emergency. Reuters News. Retrieved from http://www.reuters.com/article/2011/02/24/us-algeria-emergency-lifting-idUSTRE71N6VS20110224.
Reuters. (2005, April 11). Algerian hardline rebels try to scupper amnesty. Reuters News.
Sirri, Odai. (2004, October 23). Algerian opposition leader says open to peace bid. Reuters. Rogers, Steven. (2013). Beyond the Abu Sayyaf: the lessons of failure in the Philippines. Foreign Affairs, 83, (1), 15-20.
Ryan, J. (2007). The four p-words of militant Islamist radicalization and recruitment: persecution, precedent, piety, and perseverance. Studies in Conflict & Terrorism, 30, 985–1011.
148
Richardson, Louise. (2007). What Terrorists Want: Understanding the Enemy, Containing the Threat. New York: Random House.
Sageman, Marc. (2008). A Strategy for Fighting International Islamist Terrorists. Annals of the American Academy of Political and Social Science, 618, 223-231.
Sambanis, Nicholas. (2004). What is civil war? conceptual and empirical complexities of an operational definition. The Journal of Conflict Resolution, 48 (6), 814-858.
Sandler, Todd, and Enders, Walter. (2006). Distribution of transnational terrorism among countries by income classes and geography after 9/11. International Studies Quarterly, 367-393.
Sandler, Todd, and Siqueria, Kevin. (2006). Global terrorism: deterrence or pre-emption. The Canadian Journal of Economics, 39, (4), 1370-1387.
Schelling, Thomas. (1966). The Strategy of Conflict. Cambridge, Massacheuts: Harvard University Press.
Schmid, A.P., and Price, E. (2011). Selected literature on radicalization and de-radicalization of terrorists: monographs, edited volumes, grey literature and prime articles published since the 1960s. Crime Law Social Change, 55, 337–348.
Schmitt, Eric, and Thom Shanker. (2011, August 6). After 9/11, an Era of Tinker, Tailor, Jihadist, Spy.” New York Times. Retrieved from http://www.nytimes.com/2011/08/07/-sunday-review/after-911-an-era-of-tinker-tailor-jihadist-spy.html?pagewanted=all.
Sederberg, Peter C. (1995). Conciliation as counter-terrorist strategy. Journal of Peace Research, 32 (3), 295-312.
Shughart II, William F. (2006). The analytical history of terrorism. Public Choice, 128 (1/2), 7- 39.
Suzan, Benedicte, and Jeremy Shapiro. (2003). The French experience of counter-terrorism Brookings. Retrieved from http://www.brookings.edu/search?start=1&q=shapiro.
The Center for Systematic Peace. (2013). Polity Country Report: Algeria. Retrieved from http://www.systemicpeace.org/polity/polity4.htm.
The Center for Systematic Peace. (2013). Polity Country Report: Philippines. Retrieved from http://www.systemicpeace.org/polity/polity4.htm.
149
The Stationary Office. (2000). British Terrorism Act 2000. Limited under the authority and superintendence of Carol Tullo, Controller of Her Majesty’s Stationary Office and Queen’s Printer Acts of Parliament.
The Stationary Office. (2006). British Terrorism Act 2006. Limited under the authority and superintendence of Carol Tullo, Controller of Her Majesty’s Stationary Office and Queen’s Printer Acts of Parliament.
Tracy, Paul E. (2012). Terrorism research in criminology: current topics and future prospects. Crime & Delinquency, 58, 647-662.
Trager, Robert, and D.P. Zagorcheva. (2005/2006). Deterring terrorism: it can be done. International Security, 30 (3), 87-123.
Trinquier, Roger. (1964). Modern warfare: a French view of counterinsurgency. London: Pall Mall Press, 1964.
Useem, B., and O. Clayton (2009). Radicalization of U.S. prisoners. Criminology & Public Policy, 8, (3), 561-592.
Victor Asalm, R. Karl Rethemeyer, and Ian Anderson. (1999). Big allied and dangerous (BAAD) database 1 - lethality data 1998-2005." Study of Terrorism and State Responses. Retrieved from http://dvn.iq.harvard.edu/dvn/dv/start/faces/study/StudyPage.xhtml?-globalId=hdl:1902.1/16062&studyListingIndex=0_5ad32be44d3c864c156aa37873a4.
World Bank (2012). Gender Statistics. Retrieved from http://data.worldbank.org/data- catalog/gender- statistics.